scadalink-design/lmxproxy/src/ZB.MOM.WW.LmxProxy.Host/MxAccess/MxAccessClient.Connection.cs

using System;
using System.Collections.Generic;
using System.Runtime.InteropServices;
using System.Threading;
using System.Threading.Tasks;
using ArchestrA.MxAccess;
using Serilog;
using ZB.MOM.WW.LmxProxy.Host.Domain;

namespace ZB.MOM.WW.LmxProxy.Host.MxAccess
{
    public sealed partial class MxAccessClient
    {
        /// <summary>
        ///     Connects to MxAccess on the dedicated STA thread.
        /// </summary>
        public async Task ConnectAsync(CancellationToken ct = default)
        {
            if (_disposed) throw new ObjectDisposedException(nameof(MxAccessClient));
            if (IsConnected) return;

            SetState(ConnectionState.Connecting);

            try
            {
                await _staThread.RunAsync(() => ConnectInternal());

                lock (_lock)
                {
                    _connectedSince = DateTime.UtcNow;
                }

                SetState(ConnectionState.Connected);
                Log.Information("Connected to MxAccess (handle={Handle})", _connectionHandle);

                // Recreate any stored subscriptions from a previous connection
                await RecreateStoredSubscriptionsAsync();

                // Start persistent probe subscription
                await StartProbeSubscriptionAsync();
            }
            catch (Exception ex)
            {
                Log.Error(ex, "Failed to connect to MxAccess");
                await CleanupComObjectsAsync();
                SetState(ConnectionState.Error, ex.Message);
                throw;
            }
        }

        /// <summary>
        ///     Disconnects from MxAccess on the dedicated STA thread.
        /// </summary>
        public async Task DisconnectAsync(CancellationToken ct = default)
        {
            if (!IsConnected) return;

            SetState(ConnectionState.Disconnecting);

            try
            {
                await _staThread.RunAsync(() => DisconnectInternal());

                SetState(ConnectionState.Disconnected);
                Log.Information("Disconnected from MxAccess");
            }
            catch (Exception ex)
            {
                Log.Error(ex, "Error during disconnect");
                SetState(ConnectionState.Error, ex.Message);
            }
        }

        /// <summary>
        ///     Starts the auto-reconnect monitor loop.
        ///     Call this after initial ConnectAsync succeeds.
        /// </summary>
        public void StartMonitorLoop()
        {
            if (!_autoReconnect) return;

            _reconnectCts = new CancellationTokenSource();
            Task.Run(() => MonitorConnectionAsync(_reconnectCts.Token));
        }

        /// <summary>
        ///     Stops the auto-reconnect monitor loop.
        /// </summary>
        public void StopMonitorLoop()
        {
            _reconnectCts?.Cancel();
        }

        /// <summary>Gets the UTC time when the connection was established.</summary>
        public DateTime ConnectedSince
        {
            get { lock (_lock) { return _connectedSince; } }
        }

        /// <summary>Gets the number of times the client has reconnected since startup.</summary>
        public int ReconnectCount => _reconnectCount;

        // ── Internal synchronous methods ──────────

        private void ConnectInternal()
        {
            lock (_lock)
            {
                // Create COM object
                _lmxProxy = new LMXProxyServer();

                // Wire event handlers
                _lmxProxy.OnDataChange += OnDataChange;
                _lmxProxy.OnWriteComplete += OnWriteComplete;

                // Register with MxAccess using unique client name
                _connectionHandle = _lmxProxy.Register(_clientName);
                Log.Information("Registered with MxAccess as '{ClientName}'", _clientName);

                if (_connectionHandle <= 0)
                {
                    throw new InvalidOperationException("Failed to register with MxAccess - invalid handle returned");
                }
            }
        }

        private void DisconnectInternal()
        {
            lock (_lock)
            {
                if (_lmxProxy == null || _connectionHandle <= 0) return;

                try
                {
                    // Unadvise all active subscriptions before unregistering
                    foreach (var kvp in new Dictionary<string, int>(_addressToHandle))
                    {
                        try
                        {
                            _lmxProxy.UnAdvise(_connectionHandle, kvp.Value);
                            _lmxProxy.RemoveItem(_connectionHandle, kvp.Value);
                        }
                        catch (Exception ex)
                        {
                            Log.Debug(ex, "Error removing subscription for {Address} during disconnect", kvp.Key);
                        }
                    }

                    // Remove event handlers
                    _lmxProxy.OnDataChange -= OnDataChange;
                    _lmxProxy.OnWriteComplete -= OnWriteComplete;

                    // Unregister
                    _lmxProxy.Unregister(_connectionHandle);
                }
                catch (Exception ex)
                {
                    Log.Warning(ex, "Error during MxAccess unregister");
                }
                finally
                {
                    // Force-release COM object
                    try
                    {
                        Marshal.ReleaseComObject(_lmxProxy);
                    }
                    catch { }

                    _lmxProxy = null;
                    _connectionHandle = 0;

                    // Clear handle tracking (but keep _storedSubscriptions for reconnect)
                    _handleToAddress.Clear();
                    _addressToHandle.Clear();
                    _pendingWrites.Clear();
                }
            }
        }

        /// <summary>
        ///     Subscribes to the configured probe test tag so that OnDataChange
        ///     callbacks update <see cref="_lastProbeValueTime"/>. Called after
        ///     connect (and reconnect). The subscription is stored for reconnect
        ///     replay like any other subscription.
        /// </summary>
        private async Task StartProbeSubscriptionAsync()
        {
            if (_probeTestTagAddress == null) return;

            _lastProbeValueTime = DateTime.UtcNow;

            await _staThread.RunAsync(() =>
            {
                lock (_lock)
                {
                    if (!IsConnected || _lmxProxy == null) return;

                    // Subscribe (skips if already subscribed from reconnect replay)
                    SubscribeInternal(_probeTestTagAddress);

                    // Store a no-op callback — the real work happens in OnProbeDataChange
                    // which is called from OnDataChange before the stored callback
                    _storedSubscriptions[_probeTestTagAddress] = (_, __) => { };
                }
            });

            Log.Information("Probe subscription started for {Tag} (stale threshold={ThresholdMs}ms)",
                _probeTestTagAddress, _probeStaleThresholdMs);
        }

        /// <summary>
        ///     Called from <see cref="OnDataChange"/> when a value arrives for the probe tag.
        ///     Updates the last-seen timestamp so the monitor loop can detect staleness.
        /// </summary>
        internal void OnProbeDataChange(string address, Vtq vtq)
        {
            _lastProbeValueTime = DateTime.UtcNow;
        }

        /// <summary>
        ///     Auto-reconnect monitor loop with persistent subscription probe.
        ///     - If disconnected: attempt reconnect.
        ///     - If connected and probe configured: check time since last probe value update.
        ///       If stale beyond threshold, force disconnect and reconnect.
        /// </summary>
        private async Task MonitorConnectionAsync(CancellationToken ct)
        {
            Log.Information("Connection monitor loop started (interval={IntervalMs}ms, probe={ProbeEnabled}, staleThreshold={StaleMs}ms)",
                _monitorIntervalMs, _probeTestTagAddress != null, _probeStaleThresholdMs);

            while (!ct.IsCancellationRequested)
            {
                try
                {
                    await Task.Delay(_monitorIntervalMs, ct);
                }
                catch (OperationCanceledException)
                {
                    break;
                }

                // -- Case 1: Already disconnected --
                if (!IsConnected)
                {
                    await AttemptReconnectAsync(ct);
                    // Reset probe timer so the next check gives the new connection
                    // a full interval to deliver its first OnDataChange callback
                    _lastProbeValueTime = DateTime.UtcNow;
                    continue;
                }

                // -- Case 2: Connected, no probe configured --
                if (_probeTestTagAddress == null)
                    continue;

                // -- Case 3: Connected, check probe staleness --
                var elapsed = DateTime.UtcNow - _lastProbeValueTime;
                if (elapsed.TotalMilliseconds > _probeStaleThresholdMs)
                {
                    Log.Warning("Probe tag {Tag} stale for {ElapsedMs}ms (threshold={ThresholdMs}ms) — forcing reconnect",
                        _probeTestTagAddress, (int)elapsed.TotalMilliseconds, _probeStaleThresholdMs);

                    try
                    {
                        await DisconnectAsync(ct);
                    }
                    catch (Exception ex)
                    {
                        Log.Warning(ex, "Error during forced disconnect before reconnect");
                    }

                    await AttemptReconnectAsync(ct);
                    _lastProbeValueTime = DateTime.UtcNow;
                }
            }

            Log.Information("Connection monitor loop exited");
        }

        private async Task AttemptReconnectAsync(CancellationToken ct)
        {
            Log.Information("Attempting reconnect...");
            SetState(ConnectionState.Reconnecting);

            try
            {
                await ConnectAsync(ct);
                Interlocked.Increment(ref _reconnectCount);
                Log.Information("Reconnected to MxAccess successfully (reconnect #{Count})", _reconnectCount);
            }
            catch (OperationCanceledException)
            {
                // Let the outer loop handle cancellation
            }
            catch (Exception ex)
            {
                Log.Warning(ex, "Reconnect attempt failed, will retry at next interval");
            }
        }

        /// <summary>
        ///     Cleans up COM objects on the dedicated STA thread after a failed connection.
        /// </summary>
        private async Task CleanupComObjectsAsync()
        {
            try
            {
                await _staThread.RunAsync(() =>
                {
                    lock (_lock)
                    {
                        if (_lmxProxy != null)
                        {
                            try { _lmxProxy.OnDataChange -= OnDataChange; } catch { }
                            try { _lmxProxy.OnWriteComplete -= OnWriteComplete; } catch { }
                            try { Marshal.ReleaseComObject(_lmxProxy); } catch { }
                            _lmxProxy = null;
                        }
                        _connectionHandle = 0;
                        _handleToAddress.Clear();
                        _addressToHandle.Clear();
                        _pendingWrites.Clear();
                    }
                });
            }
            catch (Exception ex)
            {
                Log.Warning(ex, "Error during COM object cleanup");
            }
        }
    }
}